Multi-parameter fusion disaster reduction platform system based on earthquake and strong vibration monitoring

Abstract

The invention discloses a multi-parameter fusion disaster reduction platform system based on earthquake and strong vibration monitoring, and the system comprises a data collection module which collects multi-source earthquake monitoring data; the data preprocessing module is used for denoising, standardizing and synchronizing the seismic waveform data; the time synchronization and space standardization module is used for ensuring time synchronization and coordinate standardization of the seismic waveform data; the local time window division module divides the waveform data into a fragment set; the dynamic time warping algorithm module is used for executing data alignment and ensuring accurate alignment of waveform fragments; the waveform feature extraction module is used for extracting waveform features; the abnormity detection module is used for identifying and marking an abnormal fluctuation mode; and the propagation path speculation module speculates the epicentral position and the propagation path according to the mark set. According to the method, accurate synchronization and alignment of multi-source earthquake monitoring data are realized, the epicenter position speculation accuracy is improved, and the method is widely applied to earthquake disaster early warning and decision support.

Description

Technical Field

[0001] This invention relates to the field of earthquake monitoring technology, and in particular to a multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring. Background Technology

[0002] With the continuous development of earthquake monitoring technology, earthquake monitoring networks around the world are becoming increasingly sophisticated, especially in providing valuable real-time information through the analysis of seismic wave propagation paths using multi-source earthquake monitoring data. However, most current earthquake monitoring systems have some shortcomings, particularly in the processing of data from multiple sensors, time synchronization, and the estimation of seismic wave propagation paths. In existing technologies, most earthquake monitoring platforms rely on measurement data from a single sensor, leading to problems such as synchronization, alignment, and noise interference when processing multi-source data. This results in low accuracy and reliability of multi-source data fusion. Furthermore, existing time synchronization algorithms and spatial coordinate normalization methods often cannot accurately process data from different types of sensors, especially when there are significant differences in seismic wave propagation velocities and characteristics, frequently leading to synchronization errors and data distortion.

[0003] While some improved Dynamic Time Warping (DTW) algorithms and seismic waveform feature extraction methods in existing technologies have improved data alignment and seismic wave feature analysis to some extent, shortcomings remain. For example, traditional DTW algorithms fail to consider frequency and amplitude variations in multi-source seismic monitoring data, making them unsuitable for situations with a wide range of seismic wave frequencies and drastic waveform changes. Furthermore, traditional anomaly detection methods cannot effectively capture subtle differences between different seismic waveforms, often failing to accurately identify different wave patterns before, during, and after an earthquake, thus affecting the accuracy of epicenter location estimation.

[0004] Therefore, existing technologies cannot fully utilize the advantages of multi-source earthquake monitoring data, and cannot accurately predict seismic wave propagation paths and locate epicenters, resulting in a significant reduction in the accuracy and timeliness of earthquake disaster early warning and decision support systems. To address these issues, this invention proposes an improved method for synchronization, alignment, feature extraction, and anomaly detection based on multi-source seismic waveform data. Through an innovative dynamic time warping algorithm and multi-level data fusion, it provides a more accurate and reliable solution for epicenter location and propagation path prediction. How to provide a multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring. This invention fully utilizes an improved dynamic time warping algorithm, waveform feature extraction, and anomaly detection technology, and details how to optimize epicenter location and propagation path prediction through multi-source data fusion. This method has the advantages of high accuracy, high robustness, and strong real-time performance, effectively improving the accuracy of seismic wave propagation path prediction and providing reliable data for earthquake disaster early warning and decision support.

[0006] According to an embodiment of the present invention, a multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring includes:

[0007] The data acquisition module is used to collect multi-source earthquake monitoring data and obtain raw earthquake waveform data;

[0008] The data preprocessing module is used to denoise, standardize, and synchronize the acquired seismic waveform data to generate preprocessed seismic waveform data.

[0009] The time synchronization and spatial standardization module is used to synchronize the seismic waveform preprocessing data in time and standardize its spatial coordinates to obtain synchronized and standardized seismic waveform data.

[0010] The local time window segmentation module is used to segment synchronously standardized seismic waveform data into local time windows to obtain a set of waveform segments;

[0011] The dynamic time warping algorithm module is used to input the waveform segment set into the improved dynamic time warping algorithm, perform multi-source data alignment, and obtain the aligned waveform segment set.

[0012] The waveform feature extraction module is used to perform waveform feature extraction on the aligned set of waveform segments to obtain a waveform feature set.

[0013] The anomaly detection module is used to perform anomaly detection based on the waveform feature set, identify and mark abnormal fluctuation patterns, and obtain an abnormal fluctuation mark set.

[0014] The propagation path inference module is used to infer the propagation path of seismic waves based on the set of anomalous wave markers, and generate data on the epicenter location and the propagation path of the seismic waves.

[0015] Optionally, modules can be integrated using the following methods:

[0016] Multi-source seismic monitoring data is collected and preprocessed to obtain seismic waveform preprocessed data;

[0017] The preprocessed seismic waveform data is synchronized in time and standardized in spatial coordinates to obtain synchronized and standardized seismic waveform data.

[0018] Local time windows are divided into synchronously standardized seismic waveform data to obtain a set of waveform segments;

[0019] The waveform segment set is input into the improved dynamic time warping algorithm to perform multi-source data alignment, resulting in an aligned waveform segment set.

[0020] Waveform feature extraction is performed on the aligned set of waveform segments to obtain a waveform feature set;

[0021] Anomaly detection is performed based on waveform feature set to identify and mark abnormal fluctuation patterns, resulting in an abnormal fluctuation mark set;

[0022] Based on the set of anomalous wave markers, the propagation path of seismic waves is inferred, and data on the epicenter location and the propagation path of seismic waves are generated.

[0023] Optionally, the multi-source seismic monitoring data includes raw seismic waveform data of vibration, acceleration, frequency, amplitude, and timestamps; the preprocessing steps include denoising, outlier detection and repair, and eliminating high-frequency noise and low-frequency interference in the raw seismic waveform data; synchronizing the timestamps of different sensors and converting the spatial coordinates to a unified standard coordinate system; performing frequency domain analysis on the synchronized and standardized seismic waveform data to extract waveform spectral features; normalizing the amplitude; and segmenting and smoothing the data according to the duration of the seismic event to obtain preprocessed seismic waveform data.

[0024] Optionally, the time synchronization and spatial coordinate normalization include:

[0025] The timestamps in the multi-source seismic monitoring data are deviated, and the timestamps of each sensor are corrected using a time difference detection algorithm to obtain time-synchronized seismic waveform data.

[0026] Based on the seismic characteristics and sampling frequency of each sensor, and considering the frequency range and amplitude fluctuation characteristics of the seismic waveform, the size and location of the time window for each sensor's data are dynamically determined, and the time-synchronized seismic waveform data is time-aligned.

[0027] Local time window comparison is performed on the aligned seismic waveform data, and the time synchronization accuracy is optimized by a weighted matching method.

[0028] The aligned seismic waveform data is spatially standardized, and geographic information system technology is used to convert the spatial coordinates of the sensors into a unified standard coordinate system to obtain seismic waveform data with standardized spatial coordinates.

[0029] The coordinate mapping parameters are adjusted based on the sensor location and seismic waveform data. The time-synchronized and spatially standardized seismic waveform data is then smoothed to remove fluctuations caused by sensor errors, external interference, or differences between devices.

[0030] Optionally, the local time window division includes:

[0031] Spectral analysis was performed on the synchronized and standardized seismic waveform data, and the dominant frequency of each waveform was extracted using Fourier transform.

[0032] Based on the time window length determined by the dominant frequency of each waveform, the synchronously standardized seismic waveform data is divided according to the time axis to obtain a set of waveform segments, and the duration of each waveform segment is matched with its dominant frequency.

[0033] Feature extraction is performed on the data within each waveform segment. Feature extraction includes extracting frequency features and amplitude features to obtain a set of spectral features for each waveform segment.

[0034] Optionally, the improved dynamic time warping algorithm includes:

[0035] The time stamp differences in multi-source seismic monitoring data are calculated. A time difference detection algorithm is used to calculate the time deviation between different sensors. By calculating the difference between the timestamps of each pair of sensors, the synchronization error of each sensor is obtained. A weighted method is used to correct the timestamps according to the error magnitude to obtain the time-synchronized seismic waveform data.

[0036] Based on the time-synchronized seismic waveform data, an adaptive time window selection method is adopted to dynamically determine the length and position of the time window for each sensor. The size of the time window is determined by the frequency characteristics of the seismic waves from each sensor.

[0037] For the time-synchronized waveform data, a local time window comparison algorithm is used to refine and align the waveform segments within each time window;

[0038] Based on the alignment results, a nonlinear path search method is used to perform global optimization on each waveform segment. During the optimization process, a dynamic programming algorithm is used to select the optimal path based on the matching degree of the waveform segments within the local time window.

[0039] Waveform feature extraction is performed on the optimized waveform data. The frequency, amplitude, and phase features of each waveform segment are extracted to generate a waveform feature set. For each waveform segment, the maximum amplitude, frequency range, and phase difference features are extracted to form a feature vector.

[0040] Based on the waveform feature set, a weighted fusion process is performed to average the feature vectors of the sensors to generate a unified feature set of multi-source earthquake monitoring data.

[0041] Anomaly detection is performed on the fused waveform feature set. A threshold-based anomaly detection algorithm is used to detect the features of each waveform segment, identify abnormal fluctuation patterns, and mark the epicenter location and seismic wave propagation path.

[0042] Optionally, the waveform feature extraction step includes:

[0043] Based on the synchronized and aligned seismic waveform data, the maximum amplitude, root mean square value and kurtosis of each waveform segment are extracted to generate time-domain features.

[0044] Perform a Fast Fourier Transform on each waveform segment to obtain the spectrum, extract the main frequency, the center frequency of the spectrum, and the spectral width, and generate frequency domain features;

[0045] The phase difference and phase stability are calculated for each waveform segment to obtain the phase characteristics;

[0046] By fusing time-domain features, frequency-domain features, and phase features, a comprehensive feature vector of the waveform segment is obtained.

[0047] Optionally, the anomaly detection based on the waveform feature set includes:

[0048] Time-domain, frequency-domain, and phase features are extracted from the comprehensive feature vector of the waveform segment to generate a feature vector set of the waveform. The feature vector set is then standardized by removing outliers and normalizing.

[0049] Based on the standardized feature vector set, a statistical anomaly detection method is used for analysis. The mean and standard deviation of each feature are calculated, and the Z value of each waveform segment is calculated based on the mean and standard deviation. The Z value is specifically the multiple of the standard deviation of the feature value from the mean.

[0050] Set a threshold and determine whether the waveform segment is an abnormal fluctuation pattern based on the Z value. When the Z value exceeds the set threshold, the waveform segment is considered to be an abnormal fluctuation.

[0051] The identified abnormal fluctuation patterns are marked to generate an abnormal fluctuation mark set, which includes the timestamp of the waveform segment, waveform features, and mark category.

[0052] Optionally, the step of inferring the seismic wave propagation path based on the abnormal wave marker set to generate epicenter location and seismic wave propagation path data specifically includes:

[0053] Calculate the propagation time difference of each waveform segment based on the timestamp and spatial location of each waveform segment in the abnormal fluctuation marker set;

[0054] By combining the propagation time difference with the known propagation velocity of the seismic wave, the propagation path of each waveform segment can be predicted;

[0055] Based on the propagation path of each waveform segment, the epicenter location is calculated in reverse. By tracing back the timestamp and propagation path of each waveform segment, the preliminary coordinates of the epicenter location are inferred.

[0056] The propagation paths of all waveform segments are weighted and optimized, and the propagation paths of different sensors are fused using the weighted least squares method to obtain the optimal epicenter coordinates.

[0057] Based on the calculated epicenter location and propagation path, seismic wave propagation path data is generated, including propagation direction, speed, and affected area.

[0058] The beneficial effects of this invention are:

[0059] (1) Improve the time synchronization accuracy of multi-source earthquake monitoring data: The present invention adopts an improved time synchronization algorithm, which dynamically adjusts the size and position of the time window based on the seismic characteristics of different sensors, to ensure that the seismic waveform data of different sensors can be accurately aligned, greatly reducing the time synchronization error and improving the fusion accuracy of multi-source earthquake data.

[0060] (2) Optimize the accuracy of seismic wave propagation path prediction: Through the improved dynamic time warping algorithm, the present invention can perform accurate waveform alignment based on multi-source seismic waveform data, and optimize the propagation path prediction by combining nonlinear path search method, thereby reducing the error in traditional methods and improving the accuracy of epicenter location and propagation path.

[0061] (3) Enhanced detection capability of abnormal fluctuation patterns: Based on waveform feature extraction, this invention combines threshold detection and pattern recognition technology to accurately identify and mark abnormal fluctuation patterns, thereby effectively capturing fluctuation changes before, during and after earthquakes, providing reliable data support for earthquake disaster early warning and decision support. Attached Figure Description

[0062] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0063] Figure 1 This is a modular structure diagram of the multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring proposed in this invention;

[0064] Figure 2 This is a structural diagram of the improved dynamic time warping algorithm module of the multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring proposed in this invention.

[0065] Figure 3This is a schematic diagram illustrating the working principle of the anomaly detection module in the multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring proposed in this invention. Detailed Implementation

[0066] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0067] refer to Figure 1-3 A multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring includes:

[0068] The data acquisition module is used to collect multi-source earthquake monitoring data and obtain raw earthquake waveform data;

[0069] The data preprocessing module is used to denoise, standardize, and synchronize the acquired seismic waveform data to generate preprocessed seismic waveform data.

[0070] The time synchronization and spatial standardization module is used to synchronize the seismic waveform preprocessing data in time and standardize its spatial coordinates to obtain synchronized and standardized seismic waveform data.

[0071] The local time window segmentation module is used to segment synchronously standardized seismic waveform data into local time windows to obtain a set of waveform segments;

[0072] The dynamic time warping algorithm module is used to input the waveform segment set into the improved dynamic time warping algorithm, perform multi-source data alignment, and obtain the aligned waveform segment set.

[0073] The waveform feature extraction module is used to perform waveform feature extraction on the aligned set of waveform segments to obtain a waveform feature set.

[0074] The anomaly detection module is used to perform anomaly detection based on the waveform feature set, identify and mark abnormal fluctuation patterns, and obtain an abnormal fluctuation mark set.

[0075] The propagation path inference module is used to infer the propagation path of seismic waves based on the set of anomalous wave markers, and generate data on the epicenter location and the propagation path of the seismic waves.

[0076] In this embodiment, the modules are interconnected using the following method:

[0077] Multi-source seismic monitoring data is collected and preprocessed to obtain seismic waveform preprocessed data;

[0078] The preprocessed seismic waveform data is synchronized in time and standardized in spatial coordinates to obtain synchronized and standardized seismic waveform data.

[0079] Local time windows are divided into synchronously standardized seismic waveform data to obtain a set of waveform segments;

[0080] The waveform segment set is input into the improved dynamic time warping algorithm to perform multi-source data alignment, resulting in an aligned waveform segment set. The improved dynamic time warping algorithm adaptively adjusts the size and position of the time window to ensure that seismic waveform data from different sensors can be accurately aligned, and optimizes the data matching path through local time window comparison to reduce time synchronization errors. Furthermore, it uses a nonlinear time alignment method to accurately align waveform segments, further eliminating alignment errors and improving data matching accuracy.

[0081] Waveform feature extraction is performed on the aligned set of waveform segments to obtain a waveform feature set;

[0082] Anomaly detection is performed based on waveform feature set to identify and mark abnormal fluctuation patterns, resulting in an abnormal fluctuation mark set. The abnormal fluctuation patterns are identified through threshold detection and pattern recognition technology to capture abnormal changes in earthquake waveforms, such as changes in pre-earthquake, epicenter and post-earthquake fluctuations.

[0083] Based on the set of anomalous fluctuation markers, the propagation path of seismic waves is inferred, generating data on the epicenter location and the propagation path of seismic waves. The epicenter location is inferred in real time by weighted analysis of the set of anomalous fluctuation markers and data from various sensors, combined with the propagation characteristics of seismic waves, providing accurate information on seismic activity to support disaster reduction decisions.

[0084] In this embodiment, the multi-source seismic monitoring data includes raw seismic waveform data with vibration, acceleration, frequency, amplitude, and timestamps. The preprocessing steps include denoising, outlier detection and repair, and eliminating high-frequency noise and low-frequency interference in the raw seismic waveform data; synchronizing the timestamps of different sensors and converting the spatial coordinates to a unified standard coordinate system; performing frequency domain analysis on the synchronized and standardized seismic waveform data to extract waveform spectral features; normalizing the amplitude; and segmenting and smoothing the data according to the duration of the seismic event to obtain preprocessed seismic waveform data.

[0085] In this embodiment, the time synchronization and spatial coordinate standardization include:

[0086] The timestamps in the multi-source seismic monitoring data are deviated, and the timestamps of each sensor are corrected using a time difference detection algorithm to obtain time-synchronized seismic waveform data.

[0087] A time difference detection algorithm is used for time synchronization of acquired multi-source seismic monitoring data. The algorithm first detects the time deviation between data from different sensors by calculating the timestamp differences. Specifically, the algorithm compares the timestamp differences of each sensor and uses the standard deviation method to identify significant time deviations. For detected time deviations, the algorithm uses weighted averaging or linear interpolation to correct the timestamps, aligning the data from different sensors to a unified time axis. The algorithm further reduces errors in the time synchronization process by optimizing the timestamp differences of all sensors using the least squares method, thus achieving high-precision time synchronization. This algorithm ensures the temporal consistency of multi-source seismic monitoring data, providing an accurate time reference for waveform alignment and analysis.

[0088] Based on the seismic characteristics and sampling frequency of each sensor, and considering the frequency range and amplitude fluctuation characteristics of the seismic waveform, the algorithm dynamically determines the size and location of the time window for each sensor's data, and performs time alignment on the time-synchronized seismic waveform data. Specifically, the process of dynamically determining the size and location of the time window is based on the analysis of seismic waveform characteristics, including frequency range and amplitude fluctuation. The algorithm extracts the frequency components and amplitude changes of each sensor's waveform through frequency domain analysis. If the frequency in the seismic waveform is high, the algorithm selects a smaller time window to finely align the high-frequency components; if the frequency is low or the amplitude is large, the algorithm selects a larger time window to capture seismic wave changes over a longer period.

[0089] In this way, the position of the time window is dynamically adjusted according to the start point, peak time and end time of the seismic wave. For areas with large seismic wave variations, the time window is adjusted to the start and end points of the wave to ensure time synchronization accuracy. Within each time window, the algorithm uses a weighted matching method to optimize the alignment path and further refine the size and position of the time window to ensure optimal alignment of multi-source seismic data. Finally, through iterative adjustments, the optimal size and position of the time window are obtained, so that the data from all sensors can be aligned to the greatest extent on the time axis.

[0090] Local time window comparison is performed on the aligned seismic waveform data, and the time synchronization accuracy is optimized by a weighted matching method.

[0091] The aligned seismic waveform data is spatially standardized, and geographic information system technology is used to convert the spatial coordinates of the sensors into a unified standard coordinate system to obtain seismic waveform data with standardized spatial coordinates.

[0092] The coordinate mapping parameters are adjusted based on the sensor location and seismic waveform data. The time-synchronized and spatially standardized seismic waveform data is then smoothed to remove fluctuations caused by sensor errors, external interference, or differences between devices.

[0093] In this embodiment, the local time window division includes:

[0094] Spectral analysis was performed on the synchronized and standardized seismic waveform data. Fourier transform was used to extract the dominant frequency of each waveform. Based on the dominant frequency of each waveform, the required time window length was calculated. Specifically, if the dominant frequency is between 0.1 Hz and 10 Hz, a time window of 0.1 to 0.5 seconds is selected; if the dominant frequency is between 10 Hz and 50 Hz, a time window of 0.5 to 1 second is selected; and if the dominant frequency is greater than 50 Hz, a time window of 1 to 2 seconds is selected.

[0095] Based on the time window length determined by the dominant frequency of each waveform, the synchronously standardized seismic waveform data is divided along the time axis to obtain a set of waveform segments. The duration of each waveform segment matches its dominant frequency, ensuring that each waveform segment contains the complete seismic wave cycle and avoiding the loss of key changes in the seismic wave.

[0096] Feature extraction is performed on the data within each waveform segment. Feature extraction includes extracting frequency features and amplitude features to obtain a set of spectral features for each waveform segment.

[0097] By adjusting the start and end positions of the time window, we can ensure that each waveform segment accurately reflects the complete change process of the seismic wave, avoid overlap or omission between segments, and ensure the accuracy of data segmentation.

[0098] In this embodiment, the improved dynamic time warping algorithm includes:

[0099] The time stamp differences in multi-source seismic monitoring data are calculated. A time difference detection algorithm is used to calculate the time deviation between different sensors. By calculating the difference between the timestamps of each pair of sensors, the synchronization error of each sensor is obtained. A weighted method is used to correct the timestamps according to the error magnitude to obtain the time-synchronized seismic waveform data.

[0100] Based on the time-synchronized seismic waveform data, an adaptive time window selection method is adopted to dynamically determine the length and position of the time window for each sensor. The size of the time window is determined by the frequency characteristics of the seismic waves from each sensor; higher-frequency seismic waves are selected for shorter time windows, and lower-frequency seismic waves are selected for longer time windows. The start and end points of each time window are determined by the significant changes in the seismic waves, ensuring that each time window contains complete seismic wave information.

[0101] For the time-synchronized waveform data, a local time window comparison algorithm is used to refine and align the waveform segments within each time window. This step optimizes the alignment path by calculating the waveform similarity within the local time window and using a weighted matching method to reduce synchronization errors within the local time window.

[0102] Based on the alignment results, a nonlinear path search method is used to perform global optimization on each waveform segment. During the optimization process, a dynamic programming algorithm is used to select the optimal path based on the matching degree of the waveform segments within the local time window, ensuring accurate alignment of waveform segments between different sensors.

[0103] In this embodiment, the improved dynamic time warping algorithm uses dynamic programming to optimize the alignment path of waveform segments, ensuring accurate matching of multi-source seismic monitoring data. The basic idea of ​​this algorithm is to decompose the complex alignment problem into multiple smaller problems, and avoid repeated calculations by saving intermediate calculation results, thereby improving efficiency and reducing computational complexity.

[0104] During waveform alignment, the algorithm first generates a similarity matrix by calculating the similarity between each pair of sensor waveform segments. The similarity matrix reflects the matching degree between each pair of waveform segments, and the algorithm iterates through this matrix to calculate each possible matching path from the starting point to the ending point.

[0105] The dynamic programming algorithm selects the optimal path based on the cumulative matching degree of each path. It selects the path with the minimum cumulative matching degree at each step and backtracks based on the selected path to ultimately determine the best alignment path for each waveform segment.

[0106] In this way, the dynamic programming algorithm can optimize the matching relationship between waveform segments based on their similarity, thereby obtaining the optimal alignment result. This process not only reduces local matching errors but also ensures the accuracy of global waveform alignment, enabling multi-source seismic monitoring data to be accurately aligned on the time axis.

[0107] Waveform feature extraction is performed on the optimized waveform data. The frequency, amplitude, and phase features of each waveform segment are extracted to generate a waveform feature set. For each waveform segment, the maximum amplitude, frequency range, and phase difference features are extracted to form a feature vector.

[0108] Based on the waveform feature set, a weighted fusion process is performed to average the feature vectors of the sensors to generate a unified multi-source seismic monitoring data feature set, ensuring that the data from all sensors can be effectively fused under a unified time axis and spatial coordinates.

[0109] Anomaly detection is performed on the fused waveform feature set. A threshold-based anomaly detection algorithm is used to detect the features of each waveform segment, identify abnormal fluctuation patterns, and mark the epicenter location and seismic wave propagation path. This result will be fed back into earthquake disaster early warning.

[0110] In this embodiment, the waveform feature extraction step includes:

[0111] Based on the synchronized and aligned seismic waveform data, the maximum amplitude, root mean square value and kurtosis of each waveform segment are extracted to generate time-domain features.

[0112] Perform a Fast Fourier Transform on each waveform segment to obtain the spectrum, extract the main frequency, the center frequency of the spectrum, and the spectral width, and generate frequency domain features;

[0113] The phase difference and phase stability are calculated for each waveform segment to obtain the phase characteristics;

[0114] By fusing time-domain features, frequency-domain features, and phase features, a comprehensive feature vector of the waveform segment is obtained.

[0115] In this embodiment, the anomaly detection based on waveform feature sets includes:

[0116] Time-domain, frequency-domain, and phase features are extracted from the comprehensive feature vector of the waveform segment to generate a feature vector set of the waveform. The feature vector set is then standardized by removing outliers and normalizing to ensure consistency among the features and eliminate the influence of factors such as wave amplitude differences and sensor noise.

[0117] Based on the standardized feature vector set, a statistical anomaly detection method is used for analysis. The mean and standard deviation of each feature are calculated, and the Z value of each waveform segment is calculated based on the mean and standard deviation. The Z value is specifically the multiple of the standard deviation of the feature value from the mean.

[0118] Set a threshold and determine whether the waveform segment is an abnormal fluctuation pattern based on the Z value. When the Z value exceeds the set threshold, the waveform segment is considered to be an abnormal fluctuation.

[0119] The identified abnormal fluctuation patterns are marked to generate an abnormal fluctuation mark set, which includes the timestamp of the waveform segment, waveform features, and mark category.

[0120] In this embodiment, the step of inferring the seismic wave propagation path based on the abnormal wave marker set and generating epicenter location and seismic wave propagation path data specifically includes:

[0121] Calculate the propagation time difference of each waveform segment based on the timestamp and spatial location of each waveform segment in the abnormal fluctuation marker set;

[0122] By combining the propagation time difference with the known propagation velocity of the seismic wave, the propagation path of each waveform segment can be predicted;

[0123] Based on the propagation path of each waveform segment, the epicenter location is calculated in reverse. By tracing back the timestamp and propagation path of each waveform segment, the preliminary coordinates of the epicenter location are inferred.

[0124] The propagation paths of all waveform segments are weighted and optimized, and the propagation paths of different sensors are fused using the weighted least squares method to obtain the optimal epicenter coordinates.

[0125] Based on the calculated epicenter location and propagation path, seismic wave propagation path data is generated, including propagation direction, speed, and affected area.

[0126] Specifically, the timestamp, amplitude, and frequency characteristics of each waveform segment are extracted from the abnormal fluctuation marker set to establish the propagation information of each waveform segment, forming a preliminary feature set of the waveform segment. Each waveform segment has a relative relationship with other segments in time and space, and the propagation dynamics of the seismic wave are described through these feature sets.

[0127] By calculating the propagation time difference between each waveform segment and combining the spatial location information of the waveform segments, the propagation path of the waveform segments is inferred. Using the spatial extension model, based on the propagation law of seismic waves, the propagation direction and velocity of each waveform segment are inferred. The propagation direction of each waveform segment is calculated by the timestamp and the propagation time difference of adjacent waveform segments, thus determining the path of the seismic wave propagating outward from the epicenter.

[0128] The backpropagation algorithm is used to gradually deduce the epicenter position from the earliest abnormal wave segment. The timestamp and spatial position of each waveform segment are used as input. The propagation path prediction depends on the relative position and propagation speed between sensors. When predicting the path, the propagation speed and amplitude attenuation of each waveform segment are taken into account. By combining the data with the surrounding sensors, the path prediction is optimized and the direction and speed of the seismic wave propagation are corrected.

[0129] A weighted optimization method is used to fuse waveform data from each sensor, assigning different weights to each waveform segment. The weights depend on the signal-to-noise ratio, stability, and reliability of the waveform segment. After fusion, based on the propagation path inferred from each waveform segment and combined with known seismic wave propagation characteristics, the epicenter location is corrected through spatial mapping, ultimately yielding the coordinates of the epicenter.

[0130] Using a minimum error adjustment method, multiple predicted paths and epicenter locations are fused based on data from different sensors. The most suitable propagation path and epicenter location are selected to generate epicenter coordinates and seismic wave propagation path data. The final output seismic wave propagation path data includes information such as propagation direction, velocity, and propagation range, which is used for earthquake disaster early warning and post-disaster assessment.

[0131] Example 1:

[0132] This embodiment demonstrates the application of the present invention in multi-source seismic monitoring data processing. The system addresses problems in traditional methods such as low alignment accuracy, insufficient waveform feature extraction, and inaccurate anomaly identification by synchronizing, aligning, extracting features, and detecting anomalies in multi-source seismic monitoring data. This embodiment verifies the effectiveness of the present invention using specific seismic monitoring data, showcasing the advantages of the improved dynamic time warping algorithm in seismic wave propagation path prediction.

[0133] In this scenario, multiple seismic sensors simultaneously acquire seismic waveform data from different regions. These data involve different types of seismic waves (such as P-waves and S-waves). Due to the different installation locations and sampling frequencies of the sensors, the data is asynchronous in time and space. This makes it difficult for traditional processing methods to provide high-precision synchronization alignment and propagation path inference in practical applications.

[0134] To address this issue, this invention first synchronizes and standardizes the spatial coordinates of seismic waveform data from different sensors. Through weighted averaging and interpolation, data from different sensors are aligned on a unified time axis, ensuring the accuracy of subsequent processing. Next, based on the synchronized and standardized data, an improved dynamic time warping algorithm is used to align the multi-source data. This algorithm adaptively adjusts the size and position of each time window to ensure precise alignment of seismic waves and further eliminates time synchronization errors by optimizing the data matching path through local time window comparison.

[0135] In the waveform feature extraction stage, the system extracts the time-domain, frequency-domain, and phase features of each waveform segment and uses these features for anomaly detection. Through a threshold-based anomaly detection algorithm, the system identifies multiple anomalous wave patterns. These patterns demonstrate higher recognition accuracy compared to traditional methods and can effectively capture different wave patterns before, during, and after an earthquake.

[0136] Ultimately, based on the set of anomalous wave markers, the system accurately predicted the epicenter location and calculated the propagation path of the seismic waves using a reverse inference algorithm. By optimizing the multi-source data using the weighted least squares method, the system generated accurate epicenter location and propagation path data, providing reliable support for earthquake disaster early warning and decision-making.

[0137] To verify the effectiveness of this invention, we used seismic waveform data collected by multiple sensors in a certain location as test samples. This data contained seismic wave data from multiple seismic events, and each sensor had significant time synchronization and spatial errors. After processing by this invention, the data synchronization accuracy and the accuracy of propagation path prediction were significantly improved. Specific data are as follows:

[0138] Table 1: Comparison of seismic waveform data before and after processing

[0139] Data source Synchronization error before processing (seconds) Synchronization error after processing (seconds) Error in epicenter location estimation (km) Waveform feature extraction accuracy Abnormal fluctuation identification accuracy Sensor A 0.032 0.002 3.5 88% 95% Sensor B 0.045 0.003 3.2 85% 92% Sensor C 0.027 0.001 3.1 90% 97% Sensor D 0.038 0.004 3.6 87% 94% Sensor E 0.042 0.003 3.3 89% 96%

[0140] Based on the data analysis in Table 1 above, it can be seen that in processing multi-source earthquake monitoring data, this invention improves the accuracy of time synchronization and epicenter location estimation. Before processing, the synchronization error shows differences in data synchronization among the sensors, with a maximum of 0.045 seconds and a minimum of 0.027 seconds, indicating significant errors in time alignment between different sensors. However, after processing, the synchronization error of all sensors is significantly reduced after the time synchronization and spatial standardization processing of this invention, with a maximum of 0.004 seconds and a minimum of 0.001 seconds, verifying the high efficiency of this invention in multi-source earthquake data synchronization.

[0141] Regarding the error in epicenter location estimation, after data processing according to this invention, the estimation errors of all sensors are between 3.1 km and 3.6 km, far lower than the error range of traditional methods. This indicates that this invention improves the accuracy of epicenter location estimation and can pinpoint the epicenter more precisely.

[0142] Regarding waveform feature extraction accuracy, all sensors maintained an extraction accuracy between 85% and 90%, demonstrating the efficiency and stability of this invention in extracting seismic waveform features (such as amplitude and frequency). Sensor C, in particular, achieved a waveform feature extraction accuracy of 90%, validating the algorithm's advantages in high-frequency seismic wave data processing.

[0143] The accuracy of abnormal wave identification also showed a significant improvement, with all sensors achieving an accuracy between 92% and 97%, indicating that the present invention demonstrates high accuracy in identifying abnormal patterns in seismic waves (such as changes in pre-seismic, epicentral, and post-seismic waves). Sensor C achieved an anomaly identification accuracy of 97%, demonstrating the algorithm's advantages in anomaly detection, especially in accurately capturing abnormal changes in seismic waves when processing their changing patterns.

[0144] In summary, this invention significantly improves the accuracy of multi-source earthquake monitoring data processing through improved data synchronization, waveform alignment, feature extraction, and anomaly detection methods. This technology not only enhances the accuracy of epicenter location estimation but also strengthens the capabilities of waveform feature extraction and anomalous wave pattern recognition, providing a reliable data foundation for earthquake disaster early warning and subsequent decision support.

[0145] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring, characterized in that, include: The data acquisition module is used to collect multi-source earthquake monitoring data and obtain raw earthquake waveform data; The data preprocessing module is used to denoise, standardize, and synchronize the acquired seismic waveform data to generate preprocessed seismic waveform data. The time synchronization and spatial standardization module is used to synchronize the seismic waveform preprocessing data in time and standardize its spatial coordinates to obtain synchronized and standardized seismic waveform data. The local time window segmentation module is used to segment synchronously standardized seismic waveform data into local time windows to obtain a set of waveform segments; The dynamic time warping algorithm module is used to input the waveform segment set into the improved dynamic time warping algorithm, perform multi-source data alignment, and obtain the aligned waveform segment set. The waveform feature extraction module is used to perform waveform feature extraction on the aligned set of waveform segments to obtain a waveform feature set. The anomaly detection module is used to perform anomaly detection based on the waveform feature set, identify and mark abnormal fluctuation patterns, and obtain an abnormal fluctuation mark set. The propagation path inference module is used to infer the propagation path of seismic waves based on the set of anomalous wave markers, and generate data on the epicenter location and the propagation path of the seismic waves.

2. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 1, characterized in that, The modules are connected in the following way: Multi-source seismic monitoring data is collected and preprocessed to obtain seismic waveform preprocessed data; The preprocessed seismic waveform data is synchronized in time and standardized in spatial coordinates to obtain synchronized and standardized seismic waveform data. Local time windows are divided into synchronously standardized seismic waveform data to obtain a set of waveform segments; The waveform segment set is input into the improved dynamic time warping algorithm to perform multi-source data alignment, resulting in an aligned waveform segment set. Waveform feature extraction is performed on the aligned set of waveform segments to obtain a waveform feature set; Anomaly detection is performed based on waveform feature set to identify and mark abnormal fluctuation patterns, resulting in an abnormal fluctuation mark set; Based on the set of anomalous wave markers, the propagation path of seismic waves is inferred, and data on the epicenter location and the propagation path of seismic waves are generated.

3. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 2, characterized in that, The multi-source seismic monitoring data includes raw seismic waveform data with vibration, acceleration, frequency, amplitude, and timestamp; the preprocessing steps include denoising, outlier detection and repair, and elimination of high-frequency noise and low-frequency interference on the raw seismic waveform data. The timestamps of different sensors are synchronized, and the spatial coordinates are converted into a unified standard coordinate system. Frequency domain analysis is performed on the synchronized and standardized seismic waveform data to extract waveform spectral features. The amplitude is normalized, and the data is segmented and smoothed according to the duration of the seismic event to obtain preprocessed seismic waveform data.

4. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 3, characterized in that, The time synchronization and spatial coordinate standardization include: The timestamps in the multi-source seismic monitoring data are deviated, and the timestamps of each sensor are corrected using a time difference detection algorithm to obtain time-synchronized seismic waveform data. Based on the seismic characteristics and sampling frequency of each sensor, and considering the frequency range and amplitude fluctuation characteristics of the seismic waveform, the size and location of the time window for each sensor's data are dynamically determined, and the time-synchronized seismic waveform data is time-aligned. Local time window comparison is performed on the aligned seismic waveform data, and the time synchronization accuracy is optimized by a weighted matching method. The aligned seismic waveform data is spatially standardized, and geographic information system technology is used to convert the spatial coordinates of the sensors into a unified standard coordinate system to obtain seismic waveform data with standardized spatial coordinates. The coordinate mapping parameters are adjusted based on the sensor location and seismic waveform data. The time-synchronized and spatially standardized seismic waveform data is then smoothed to remove fluctuations caused by sensor errors, external interference, or differences between devices.

5. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 4, characterized in that, The local time window division includes: Spectral analysis was performed on the synchronized and standardized seismic waveform data, and the dominant frequency of each waveform was extracted using Fourier transform. Based on the time window length determined by the dominant frequency of each waveform, the synchronously standardized seismic waveform data is divided according to the time axis to obtain a set of waveform segments, and the duration of each waveform segment is matched with its dominant frequency. Feature extraction is performed on the data within each waveform segment. Feature extraction includes extracting frequency features and amplitude features to obtain a set of spectral features for each waveform segment.

6. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 5, characterized in that, The improved dynamic time warping algorithm includes: The time stamp differences in multi-source seismic monitoring data are calculated. A time difference detection algorithm is used to calculate the time deviation between different sensors. By calculating the difference between the timestamps of each pair of sensors, the synchronization error of each sensor is obtained. A weighted method is used to correct the timestamps according to the error magnitude to obtain the time-synchronized seismic waveform data. Based on the time-synchronized seismic waveform data, an adaptive time window selection method is adopted to dynamically determine the length and position of the time window for each sensor. The size of the time window is determined by the frequency characteristics of the seismic waves from each sensor. For the time-synchronized waveform data, a local time window comparison algorithm is used to refine and align the waveform segments within each time window; Based on the alignment results, a nonlinear path search method is used to perform global optimization on each waveform segment. During the optimization process, a dynamic programming algorithm is used to select the optimal path based on the matching degree of the waveform segments within the local time window. Waveform feature extraction is performed on the optimized waveform data. The frequency, amplitude, and phase features of each waveform segment are extracted to generate a waveform feature set. For each waveform segment, the maximum amplitude, frequency range, and phase difference features are extracted to form a feature vector. Based on the waveform feature set, a weighted fusion process is performed to average the feature vectors of the sensors to generate a unified feature set of multi-source earthquake monitoring data. Anomaly detection is performed on the fused waveform feature set. A threshold-based anomaly detection algorithm is used to detect the features of each waveform segment, identify abnormal fluctuation patterns, and mark the epicenter location and seismic wave propagation path.

7. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 6, characterized in that, The waveform feature extraction step includes: Based on the synchronized and aligned seismic waveform data, the maximum amplitude, root mean square value and kurtosis of each waveform segment are extracted to generate time-domain features. Perform a Fast Fourier Transform on each waveform segment to obtain the spectrum, extract the main frequency, the center frequency of the spectrum, and the spectral width, and generate frequency domain features; The phase difference and phase stability are calculated for each waveform segment to obtain the phase characteristics; By fusing time-domain features, frequency-domain features, and phase features, a comprehensive feature vector of the waveform segment is obtained.

8. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 7, characterized in that, The anomaly detection based on waveform feature sets includes: Time-domain, frequency-domain, and phase features are extracted from the comprehensive feature vector of the waveform segment to generate a feature vector set of the waveform. The feature vector set is then standardized by removing outliers and normalizing. Based on the standardized feature vector set, a statistical anomaly detection method is used for analysis. The mean and standard deviation of each feature are calculated, and the Z value of each waveform segment is calculated based on the mean and standard deviation. The Z value is specifically the standard deviation multiple of the feature value from the mean. Set a threshold and determine whether the waveform segment is an abnormal fluctuation pattern based on the Z value. When the Z value exceeds the set threshold, the waveform segment is considered to be an abnormal fluctuation. The identified abnormal fluctuation patterns are marked to generate an abnormal fluctuation mark set, which includes the timestamp of the waveform segment, waveform features, and mark category.

9. The multi-parameter fusion disaster reduction platform system based on earthquake and strong ground motion monitoring according to claim 8, characterized in that, The step of inferring the seismic wave propagation path based on the abnormal wave marker set to generate epicenter location and seismic wave propagation path data specifically includes: Calculate the propagation time difference of each waveform segment based on the timestamp and spatial location of each waveform segment in the abnormal fluctuation marker set; By combining the propagation time difference with the known propagation velocity of the seismic wave, the propagation path of each waveform segment can be predicted; Based on the propagation path of each waveform segment, the epicenter location is calculated in reverse. By tracing back the timestamp and propagation path of each waveform segment, the preliminary coordinates of the epicenter location are inferred. The propagation paths of all waveform segments are weighted and optimized, and the propagation paths of different sensors are fused using the weighted least squares method to obtain the optimal epicenter coordinates. Based on the calculated epicenter location and propagation path, seismic wave propagation path data is generated, including propagation direction, speed, and affected area.

Images